The recognition of the start codon by the eukaryotic translational machinery is a critical step in gene expression. It sets the reading frame in which the mRNA will be decoded and defines the N-terminus of the corresponding protein. We have developed an in vivo high-throughput dual luciferase assay that allows accurate measurement of the fidelity of start codon recognition in S. cerevisiae. Using this assay and a powerful secondary assay that eliminates agents that generally affect expression or activity of one or both luciferase enzyme, we screened over 50,000 molecules for compounds that reduce the fidelity of start codon recognition. From this screen two structurally similar molecules were found that increase use of near-cognate codons (e.g., UUG, CUG, etc.) as initiation sites. These activities were confirmed in several tertiary assays. Further in vivo analysis of the effects of the compounds supported the hypothesis that they act directly on the translational apparatus. These are the first compounds to be identified that modulate the fidelity of start codon recognition. We propose to continue this screen to find additional compounds with stronger effects on the fidelity of start codon recognition. Collaboration with one of the MLPCN centers will dramatically increase the efficiency of this undertaking. In addition, the chemical synthesis expertise of the centers will enable detailed structure-activity relationships to be performed, as well as optimization of the efficacy of active compounds that are discovered. In addition to their inherent utility as tools for probing the molecular mechanics of start codon recognition, compounds that modulate the fidelity of this process would be valuable leads for new antiproliferative agents, antifungal and antiparasitic agents, and therapeutics to ameliorate genetic diseases caused by initiation codon mutations.